When Atalmedial met AI Startup Lab

In what’s truly a match made in healthcare innovation heaven, one of the largest medical diagnostics laboratories in the Netherlands, Atalmedial, is currently collaborating with AI Startup Lab (an initiative by ACE) to streamline anaemia treatments. As the company explores the potential of AI to bring more value to their clients, the young student team confronts a real-world problem using a comprehensive anonymised data set. “This access to so much data is rare,” explains one of the AI team. “So, if there’s a signal there, we’ll find it.”

Inspiring the health and life sciences sector

“We’re not one of these large corporates looking to infuse our culture with a lean and agile startup mentality. Being in the healthcare industry, we have other challenges to address,” says Ivo Kerklaan, the innovation manager at Atalmedial. “Seeing the potential of AI in healthcare, we have already done some work with AI and we found out it’s best approached as a collaboration. We’re already domain experts, so let’s call in the AI experts.”

That’s when Jeroen Maas, health lead at Amsterdam Economic Board and an initiator behind Smart Health Amsterdam, played matchmaker.

Atalmedial is currently in the middle of a 12-week collaboration with a team put together by AI Startup Lab, an initiative of the Amsterdam Center for Entrepreneurship (ACE). The idea is to connect companies that have AI ambitions with AI students who are craving real-world experience using a relatively clean data set. Everyone wins.

“It’s a really unique proposition,” says Andrew Harrison, who is a master’s student of Artificial Intelligence at the University of Amsterdam and a member of the team working on the challenge. “Instead of having to first go in search of a problem, you’re given a challenge that needs solving. There’s a whole support network you can tap into. And certainly, this level of access to data is rare. Plus, you get to explore all this data with a group of smart people. There are so many shortcuts!”

Acting now on AI ambitions

Atalmedial is active in medical diagnostics with a strong side-line in thrombosis treatment. “Basically, we do everything from the drawing of a blood sample to the sending of results,” says Kerklaan.

“With AI, we hope to bring added value to patients and healthcare. But we realised early on that AI development is not our core ability. At the same time, we don’t want to wait for third parties. We also don’t have the resources to set up large-scale academic projects or pay a commercial AI development company by the hour. With this project we can start now.”

AI is a natural fit for Atalmedial

Atalmedial deals with 480,000 patients a year – which comes down to 3 million tubes of blood and almost 15 million test results. “We really do have a lot of data points, so applying AI is a natural progression for us,” explains Bauke de Boer. As the company’s Laboratory Specialist, he’s working most closely with the startup team.

“Yes, we are most efficient in providing our diagnostic results,” says De Boer. “But how can we make more of a contribution to a patient’s health and the healthcare process? To help those in the field make better informed decisions? How can we take a more advisory role based on diagnostics? Right now, we just give GPs a number – a coded message. How can we turn this data into useable information?”

“The AI Startup Lab seems to fit the DIY part of our AI strategy perfectly,” says Kerklaan. “We supply anonymous data and our domain knowledge without the usual big financial investment. The startup approach provides strong applicational and viability focus. And any successful results can be brought back to benefit our patients and healthcare providers.”

Long-term ambitions in AI

Over the long-term, De Boer sees a great deal of potential in Atalmedial using AI to personalise treatments: “We have our result ‘norms’, which are calculated from a large cohort of healthy individuals, but actually, each patient has their own norms,” he says. “I also see huge benefits if we are better enabled to predict the progression of certain diseases so we can take actions much earlier.”

In the case of the AI Startup Lab challenge, the focus is on streamlining medical decision-making: using AI to predict the most likely cause of anaemia in a particular patient, with anaemia being one of the most frequently encountered health issues in GP healthcare.

What caused this case of anaemia?

Anaemia is a decrease of haemoglobin in the blood. As a result, blood is less able to carry oxygen around the body. Symptoms range from tiredness to loss of consciousness. It has a broad range of causes, from blood loss to genetic conditions. As the most common blood disorder, affecting about a third of the global population, it’s costly in terms of treatment, productivity and quality of life.

“The cause is usually iron deficiency, but why test for it if there’s a more likely cause in a particular patient?” observes Harrison. “So, we’re looking at the data exhaust – all that extra information around a blood test – to see if we can find signals that can reliably predict the real cause. We can then apply these signals into a decision support system for GPs. The benefits would be many: less tests and therefore less costs, a faster answer, and a better customer journey for patients.”

Using health data for the power of good

While still only in the first month of the trajectory, Harrison and his team are confident that if such signals exist, they’ll find it. “The data set is amazing. We have access to two million tests taken over 15 years,” says Harrison.

The team certainly inspires confidence. While Harrison hails from Australia, Maximiliane Ekert is from Germany, Mathias Parisot is from France and Rahul Kumar is from India. “We’re a very diverse bunch with different AI specialities. And very dynamic. I have a business consulting background, but that’s not turning out to be my focus. Most of us can do most things,” says Harrison.

“I’ve done a lot around privacy and ethical issues around AI so this is particularly inspiring for me,” enthuses Harrison. “Data exhaust is often used nefariously, for advertising and such. Here, it can be used for the power of good.”

Over the next two months, the team hopes to make a minimal viable product that impresses Atalmedial enough for it to become the first user of the product. “Then with our IP in hand, we can expand.”

20 May 2021

More information

Contact us



Want to read more articles like this one?
Follow us daily on LinkedIn, Twitter and sign up for our Board Update.

Read more